A gene expression phenotype in lymphocytes from Friedreich ataxia patients

Abstract

Objective: Gene expression studies in peripheral tissues from patients with neurodegenerative disorders can provide insights into disease pathogenesis, and identify potential biomarkers, an important goal of translational research in neurodegeneration. Friedreich Ataxia (FRDA) is a chronic neurodegenerative disease caused by reduced transcription of frataxin, a ubiquitously expressed protein. We studied in vitro lymphocytes from FRDA patients and carriers to identify a peripheral gene expression phenotype. Peripheral biomarkers related to disease status would be extremely valuable for assessing drug efficacy and could provide new pathophysiological insights.

Methods: We characterized the gene expression profiles in peripheral blood mononuclear cells (PBMCs) from FRDA patients, compared with controls and related carriers. Cells were studied both before and after in vitro treatment with compounds that increase frataxin levels. Quantitative real-time polymerase chain reaction and additional microarrays were used to confirm a core set of genes in multiple independent series.

Results: We identified a subset of genes changed in cells from patients with pathological frataxin deficiency, and a core set of these genes were confirmed in independent series. Changes in gene expression were related to the mitochondria, lipid metabolism, cell cycle, and DNA repair, consistent with FRDA's known pathophysiology. We evaluated the in vitro effect of multiple compounds (histone deacetylase inhibitors) on this putative biomarker set, and found that this biochemical phenotype was ameliorated in accordance with drug efficacy.

Interpretation: Frataxin downregulation is associated with robust changes in gene expression in PBMCs, providing pathogenetic insights and a core subset of genes that, if verified in vivo, could be used as a peripheral biomarker.

Figure 1. Study design

Peripheral blood was extracted from 10 FRDA patients, 10 related heterozygous carriers, and 11 unrelated controls, for a total of 31 individuals. PBMCs were extracted and cultured for 48 hours. After RNA extraction, total RNA was amplified, labeled and hybridized on Illumina Human RefSeq-8 microarrays, querying the expression of >22,000 RefSeq-curated transcripts. Data analysis aimed at comparing FRDA patients vs. normal controls and vs. related carriers.

Figure 2. Transcriptional changes associated with frataxin deficiency

(A) Barplot representing the number of downregulated (green) and upregulated (red) genes in two comparisons: FRDA vs. controls, and Carriers vs. controls (complete gene lists are in Tables S2 and S3). (B) Venn diagram representing the number of genes shared between the two comparisons. 590 genes are DE in both FRDA and carriers vs. controls, all of them changing in the same direction. (C) Overrepresented gene ontology (GO) categories among DE genes in FRDA vs. controls (in green the proportion of downregulated DE genes; in red the proportion of upregulated) sorted by −Log₁₀ (p-value). A −Log (p-value) of 1.3 corresponds to an overrepresentation p-value of 0.05. Gene Ontology list is in Table S4. (D) Heatmap depicting fold changes of FRDA patients compared to related carriers (pairs). 77 probes are differentially expressed (p<0.005 Bayesian t-test). Upregulated genes are shown in red and downregulated genes in green; color intensity corresponds to fold-change in expression (probes are listed in Table 1). (E) Overrepresented gene ontology (GO) categories among DE genes in FRDA vs. related carriers (in green the proportion of downregulated DE genes; in red the proportion of upregulated).

Figure 3. Real-time PCR confirmation of the putative biomarker set

Barplot representing the gene expression fold changes (absolute ratios) of 19 genes selected for confirmation, in 3 independent datasets (see Text): 4L (blue bars), 10I (green bars), and 10B (orange bars). The fold change detected by the array is represented for reference with white bars. Error bars: standard error of the mean. The red line is marking a fold change = 1 (i.e. no change).

Figure 4. Biomarker validation on an independent series

(A) Clustering of 51 samples (34 FRDA patients, 17 carriers) based on the 67 probes that are overlapping with the P77 set. The two main branches classify samples with an overall 82% accuracy. Samples (Table S6) are color coded by diagnosis (dark blue: FRDA, light blue: carrier), family (subjects from the same family have the same color, except gray=no family members enrolled), gender (pink: women, blue: men), array batch (arrays from the same batch have the same color), and age category (red: >25, green: <25 years old). (B) Heatmap representing the gene expression changes in the 67 probes that are overlapping with the P77 set. Genes and samples are clustered by similarity. Red denotes upregulation and green downregulation when comparing patients vs. carriers. Fold changes in the original 10 patient/carrier pairs are reported for reference (top color code: blue). Sample numbering is as in Table S6. 42% (in patient #32) to 69% (in patient #26) of the 67 probes show a change similar to the original biomarker set.

Figure 5. Drug effect on frataxin levels and FRDA signature

(A) Heatmap representing the gene expression changes in the P77 probes in 10 patient/carrier pairs (top color code: purple) and the changes induced in these same genes by treatment with compound 106 (top color code: cyan). Red denotes upregulation and green downregulation when comparing patients vs. carriers (purple columns) or treated vs. untreated FRDA cells (cyan columns). Opposite colors in the two sections of the heatmap correspond to amelioration of the biochemical phenotype. (B) Barplot representing the percentage of the P77 genes changed towards normal (yellow) or completely normalized (red) by treatment. Trend towards normalization is defined as a change of at least 30% towards normal levels. An effect is detectable in both patient and carrier samples, for compounds 4b and 106. Compound 106 has the highest percentage of trend towards normalization (~80%) and complete normalization (~60%). (C) Frataxin mRNA levels in PBMCs after treatment with compound 106 at increasing doses (1, 5, 10, 20, and 50 μM) as detected by Illumina microarray probes show a dose-dependent increase spanning 2 log2-scale units in normal individuals (red lines), carriers (light colors) and patients (dark colors). The highest treatment dose (50 μM) is associated with a drop of frataxin levels, possibly because of toxicity; (D) Barplot representing the percentage of genes changed towards normal (yellow) or completely normalized (red) by treatment, in cells from patients and carriers, after treatment with increasing doses of c106. The trend towards normalization plateaus at 5 μM, the percentage of completely normalized genes at 10 μM; (E) qPCR confirmation of the treatment effect in an independent series. Barplot representing the gene expression fold changes (absolute ratios) of 8 genes selected for confirmation, in the 10B series (see Text): bars represent gene changes in cells from FRDA patients vs. carriers (blue bars), and effect of c106 treatment on FRDA cells. (orange bars). The fold change detected by the array is represented for reference with white bars. Error bars: standard error of the mean. The red line is marking a fold change = 1 (no change). Five out of seven genes (besides frataxin) show a reverse change after treatment with c106. For one (APOE) the drug related but not the FRDA-related change was confirmed by qPCR.